挖掘机作为土方工程机械，以其高适应性、高性价比的优点广泛应用于国民经济建设的各个领域。传统挖掘机工作装置（工装）系统运行存在时间迟滞性和控制精度差的问题，阻碍了挖掘机自动化发展。随着机械智能控制理论的逐渐深入，挖掘机智能化控制备受关注。本文将RBF神经网络引入挖掘机工装轨迹控制进行研究，以寻求更佳的控制效果，研究工作如下：

（1）分析挖掘机工装轨迹控制的基本组成，现有存在问题与不足，制定RBF神经网络工装轨迹控制的总体研究方案；通过工作装置建模，基于ADAMS分析挖掘装载作业的基本特点及工作装置各铰接点的载荷变化。

（2）建立挖掘机工装轨迹控制系统的数学模型，选用徐工XE65D型挖掘机工作装置基本参数，确定工装系统动臂、斗杆及其他环节的传递函数，对动臂与斗杆控制系统基于Simulink建模，仿真分析其控制效果。

（3）对工装轨迹控制系统进行PID控制器设计以调节系统误差，对比分析加入PID算法的控制效果，采用神经网络智能化控制理论，分析基于神经网络PID的工装轨迹控制策略，通过MATLAB的S-Function分别对动臂及斗杆等控制系统进行BP神经网络和RBF神经网络的PID控制器设计，仿真得到系统在测试信号下的响应效果。

（4）分析遗传算法的基本特点和优化流程，引入粒子群算法位置更新原理对其改进，对RBF神经网络工装轨迹控制进行优化，并基于工装控制系统仿真平台得到系统测试信号下的响应结果；通过实验环节分析工作装置铲斗末端在水平挖掘模拟实验的控制效果，进一步验证优化后的RBF神经网络工装轨迹控制的可行性。

本文通过误差调节对挖掘机工装轨迹控制进行研究，仿真与实验结果表明：GA-RBF工装轨迹控制精度较高、鲁棒性好，控制误差较PID控制减少了10mm，精度提高77%，研究结果对挖掘机自动化发展具有一定的借鉴意义。

As a kind of earthwork machinery, excavator is widely used in various fields of national economic construction because of its high adaptability and high cost performance. There are some problems in the operation of traditional excavator working device system, such as time lag and poor control accuracy, which hinder the development of excavator automation. With the development of the theory of mechanical intelligent control, the intelligent control of excavator has attracted more and more attention. In this paper, RBF neural network is introduced into the control of excavator tooling trajectory in order to find a better control effect. The research work is as follows:

       (1) Based on the analysis of the basic components of the excavator tooling trajectory control, the existing problems and shortcomings, the overall research scheme of the RBF neural network tooling trajectory control is formulated. Through the modeling of the working device, based on ADAMS, the basic characteristics of excavation loading job and the load changes of each hinge point of the working device are analyzed.

       (2) The mathematical model of the control system of excavator tooling track is established. The basic parameters of XCMG XE65D excavator working device are selected to determine the transfer function of the boom, stick and other links of the tooling system. The control system of the boom and stick is modeled based on Simulink, and the control effect is simulated and analyzed.

       (3) The PID controller is designed to adjust the system error of the tool path control system. The control effect of adding PID algorithm is compared and analyzed. The intelligent control theory of neural network is used to analyze the tool path control strategy based on Neural Network PID. Through the S-Function file of MATLAB, the PID controller of BP neural network and RBF neural network is designed for the control system of boom and stick respectively, and the response effect of the system under the test signal is obtained by simulation.

       (4) This paper analyzes the basic characteristics and optimization process of genetic algorithm, introduces the principle of particle swarm optimization to improve it, optimizes the RBF neural network tooling trajectory control, and obtains the response results under the system test signal based on the tooling control system simulation platform. By analyzing the control effect of the bucket end of the working device in the horizontal excavation simulation experiment, the feasibility of the optimized RBF neural network tooling trajectory control is further verified.

       In this paper, the control of excavator tooling trajectory is studied by error adjustment. The simulation and experimental results show that GA-RBF tooling trajectory control has high accuracy and good robustness. The control error is reduced by 10mm compared with PID control, and the accuracy is increased by 77%. The research result has certain reference significance for the development of excavator automation.